[Intel-gfx] video: X sets brightness to zero after resume

[Hans de Goede]

Hi,

On 06/13/2014 03:28 AM, Aaron Lu wrote:
> On 06/12/2014 08:42 PM, Kalle Valo wrote:
>> Hi Aaron,
>>
>> after your commit 0e9f81d3b7c ("ACPI / video: Add systems that should
>> favour native backlight interface") I have had an regression that every
>> time after resume the display brightness has been set to zero and I need
>> to manually set it to non-zero to see something again.
>>
>> Finally I started to investigate this more closely and it seems that X
>> sets it to zero for some reason. I added a WARN_ON() and few printks to
>> brightness_store() in drivers/video/backlight/backlight.c and this is
>> what I see during resume:
>>
>> [   49.228221] ------------[ cut here ]------------
>> [   49.228229] WARNING: CPU: 1 PID: 1133 at
>> drivers/video/backlight/backlight.c:173 brightness_store+0x3c/0x120(
>> )
>> [   49.228230] Modules linked in: ctr ccm uvcvideo videobuf2_core
>> videodev videobuf2_vmalloc videobuf2_memops fu
>> se arc4 iwldvm sha256_generic kvm_intel kvm mac80211 snd_hda_codec_hdmi
>> snd_hda_codec_realtek snd_hda_codec_gene
>> ric iptable_nat nf_conntrack_ipv4 nf_defrag_ipv4 snd_hda_intel
>> snd_hda_controller nf_nat_ipv4 nf_nat snd_hda_cod
>> ec nf_conntrack snd_hwdep snd_pcm_oss iwlwifi snd_mixer_oss snd_pcm
>> ip_tables x_tables thinkpad_acpi snd_seq_dum
>> my snd_seq_oss snd_seq_midi snd_rawmidi snd_seq_midi_event snd_seq
>> cfg80211 snd_timer snd_seq_device snd bnep rfcomm bluetooth psmouse
>> soundcore nvram ehci_pci ehci_hcd rfkill wmi binfmt_misc ext4 jbd2
>> mbcache sd_mod fbcon tileblit font bitblit softcursor i915
>> drm_kms_helper drm cfbcopyarea i2c_algo_bit intel_gtt agpgart i2c_core
>> video xhci_hcd cfbimgblt cfbfillrect e1000e sdhci_pci sdhci ptp pps_core
>> mmc_core ahci libahci
>> [   49.228281] CPU: 1 PID: 1133 Comm: Xorg Tainted: G        W
>> 3.15.0-rc7-wl-ath+ #259
>> [   49.228282] Hardware name: LENOVO 2324JB2/2324JB2, BIOS G2ET82WW
>> (2.02 ) 09/11/2012
>> [   49.228283]  0000000000000000 ffffffff817e4000 ffffffff814cbee3
>> 0000000000000000
>> [   49.228285]  ffffffff810493bc ffff88020e2d1c98 ffff880213bc9f10
>> 0000000000000002
>> [   49.228287]  ffff8800c5dfc198 ffff88020d7cbf50 ffffffff81290a1c
>> ffff88020e792800
>> [   49.228288] Call Trace:
>> [   49.228293]  [<ffffffff814cbee3>] ? dump_stack+0x41/0x51
>> [   49.228296]  [<ffffffff810493bc>] ? warn_slowpath_common+0x8c/0xc0
>> [   49.228298]  [<ffffffff81290a1c>] ? brightness_store+0x3c/0x120
>> [   49.228301]  [<ffffffff811d7242>] ? kernfs_fop_write+0x112/0x170
>> [   49.228304]  [<ffffffff8116791b>] ? vfs_write+0xcb/0x1f0
>> [   49.228306]  [<ffffffff81167e20>] ? SyS_write+0x50/0xb0
>> [   49.228308]  [<ffffffff814dd312>] ? system_call_fastpath+0x16/0x1b
>> [   49.228309] ---[ end trace a1f60d12df64bdbe ]---
>> [   49.228310] brightness=0
>> [   49.228310] set brightness to 0
>>
>> And this is PID 1133:
>>
>> root      1133  0.7  0.2 131504 20192 tty7     Ss+  15:16   0:00
>> /usr/bin/X :0 -auth /var/run/lightdm/root/:0 -nolisten tcp vt7
>> -novtswitch -background none
>>
>> If I set video.use_native_backlight=0 in kernel command line the problem
>> goes away. Or if I revert your commit 0e9f81d3b7c the problem also goes
>> away. Any ideas?
> 
> Not really...
> I've added i915 people maybe they have an idea.

This is a known issue in xf86-video-intel, which gets exposed due to the
video.use_native_backlight=1 option. This should be fixed in the latest 2.99.912
xf86-video-intel release.

Regards,

Hans